michaelm@bcsaic.UUCP (michael b maxwell) (07/11/85)
OK, you linguists--here's your chance!
Consider the NP "the king and queen." This NP must be plural, as we see from
agreement with the verb:
(1) The king and queen are/ *is...
Nevertheless, the determiner must be singular:
(2) This/ *These king and queen...
Similar coordinated nouns with obligatorily singular determiners are the
following:
(3) A/ *0 husband and wife should love each other.
This/ *These boy and girl are in love.
This/ *desk and chair need to go over there.
This/ *these day and night have already lasted too long for me.
The fact that the plural determiner with the coordinated singular nouns is
definitely out would seem to imply that the structure cannot be:
(4) [ Det [ Noun "and" Noun]]
NP N
since the N formed from the coordinated Nouns would presumably be plural (as
is the resulting NP), and the examples of (2-3) with plural determiners would
therefore be acceptable (and those with singular determiners would be
unacceptable).
First question: what rules out this structure? i.e. why can't Ns be coordinated?
It would seem that verbs can be coordinated, provided their subcategorization
restrictions are the same (e.g. "I saw and heard a flying saucer"). (Or is
this some sort of gapping construction?) Likewise with adjectives ("the
yellow and green gorph"), prepositions ("I looked both under and behind the
gorph") and complementizers ("John whispered that Bill had left and Mary had
stayed"). And if you believe that nouns can have subcategorized complements,
we have "the destruction and rebuilding of the city," where the nouns both
seem to be in the scope of the subcatagorized(?) PP.
At the same time, there seem to be examples of coordinated nouns that cannot
take either a singular determiner or a plural determiner (although they
happily take a determiner which is indeterminate as to plurality). Examples:
(5) ?*This/ *These/ The book and magazine are quite old.
*This/ *These/ The idea and presentation are both quite boring.
*This/ *These/ The apple and orange are good to eat.
?*This/ *These/ The computer and screwdriver are for sale.
*This/ *These/ The story and movie are quite different.
(I would say something "semanticky" is going on here--e.g. "This apple and
orange are good to eat together" sounds much better to me.)
Second question: why the difference between the two sets--those that take
singular determiners when coordinated (3), and those that don't (5)? Is it
something to do with whether the thing represented by the coordinate nouns
somehow forms a unit (in some gestalt(?) sense)? (If so, we should expect
quite a bit of difference in acceptability judgements about the examples in
(3) and (5).) Further examples of NPs like those in (3) and (5) are welcome.
Third question: if we have indeed ruled out the structure in (4), then it
would seem the only alternative structure is the following:
(6) [ [ Det N] "and" [ N]]
NP N? N?
(I put the question marks after the N labels on the inner brackets so as to
not pin myself down on the question of the number of bar-levels in NP.) If
this is the correct structure (and I welcome alternatives), then how does
the determiner get associated with the second noun? (I take it as clear that
the second noun in the examples of (2-3) is within the scope of the determiner.)Some kind of scoping/ movement rule, like with quantifiers?
Finally, anyone know any references where this kind of NPs is (are??)
discussed?
--
Mike Maxwellfranka@mmintl.UUCP (Frank Adams) (07/18/85)
In article <177@bcsaic.UUCP> michaelm@bcsaic.UUCP (michael b maxwell) writes: > [...] >(5) ?*This/ *These/ The book and magazine are quite old. > *This/ *These/ The idea and presentation are both quite boring. > *This/ *These/ The apple and orange are good to eat. > ?*This/ *These/ The computer and screwdriver are for sale. > *This/ *These/ The story and movie are quite different. >(I would say something "semanticky" is going on here--e.g. "This apple and >orange are good to eat together" sounds much better to me.) I don't agree. This seems perfectly correct to me each of these cases. >Finally, anyone know any references where this kind of NPs is (are??) >discussed? This should read "this kind of NP is discussed".
drg@rlvd.UUCP (Duncan R. Gibson) (07/22/85)
It seems to me that in all of the examples quoted, of the form This/These x and y are ... that "this" is the appropriate word to use since the sentences imply This x is ... *and* this y is ... The exapmples given are abbreviations for *two* sentences, not one sentence containing a collective subject. (I hope I've managed to express what I mean :-)
keesan@bbncc5.UUCP (Morris M. Keesan) (07/23/85)
In article <177@bcsaic.UUCP> michaelm@bcsaic.UUCP (michael b maxwell) writes: > [...] >(5) ?*This/ *These/ The book and magazine are quite old. > *This/ *These/ The idea and presentation are both quite boring. > *This/ *These/ The apple and orange are good to eat. > ?*This/ *These/ The computer and screwdriver are for sale. > *This/ *These/ The story and movie are quite different. >(I would say something "semanticky" is going on here--e.g. "This apple and >orange are good to eat together" sounds much better to me.) I agree that the singular "this" or "the" sounds better. I think there's an elided article in each of these examples. I read these as, e.g., This book and [this] magazine are quite old. The story and [the] movie are quite different. -- Morris M. Keesan keesan@bbn-unix.ARPA {decvax,ihnp4,etc.}!bbncca!keesan
lwall@sdcrdcf.UUCP (Larry Wall) (07/25/85)
In article <177@bcsaic.UUCP> michaelm@bcsaic.UUCP (michael b maxwell) writes: > OK, you linguists--here's your chance! > Consider the NP "the king and queen." This NP must be plural, as we see from > agreement with the verb: > (1) The king and queen are/ *is... First blood. Almost anything can happen in normal language, given the proper *semantic* context. For example: Q: Did you see the king carrying the bishop around piggyback? A: The king and queen is who I saw. The singlular/plural distinction can become quite mushy in real life. This plays havoc with transformational grammarians who are infected with "binary madness", but that's the breaks. > Nevertheless, the determiner must be singular: > (2) This/ *These king and queen... This is certainly true. I think there may be some "pressure" exerted by the fact that the listener who has heard "These king and queen..." has already analyzed "king" and "queen" as nouns functioning as adjectives, as in "These king and queen mattresses...". In other words, if "these A and B" were admitted to the language there would be much more possibility of "garden pathalogical" noun phrases. > Similar coordinated nouns with obligatorily singular determiners are the > following: > (3) A/ *0 husband and wife should love each other. > This/ *These boy and girl are in love. > This/ *desk and chair need to go over there. > This/ *these day and night have already lasted too long for me. > The fact that the plural determiner with the coordinated singular nouns is > definitely out would seem to imply that the structure cannot be: > (4) [ Det [ Noun "and" Noun]] > NP N > since the N formed from the coordinated Nouns would presumably be plural (as > is the resulting NP), and the examples of (2-3) with plural determiners would > therefore be acceptable (and those with singular determiners would be > unacceptable). > > First question: what rules out this structure? i.e. why can't Ns be coordinated? > It would seem that verbs can be coordinated, provided their subcategorization > restrictions are the same (e.g. "I saw and heard a flying saucer"). (Or is > this some sort of gapping construction?) Likewise with adjectives ("the > yellow and green gorph"), prepositions ("I looked both under and behind the > gorph") and complementizers ("John whispered that Bill had left and Mary had > stayed"). And if you believe that nouns can have subcategorized complements, > we have "the destruction and rebuilding of the city," where the nouns both > seem to be in the scope of the subcatagorized(?) PP. > > At the same time, there seem to be examples of coordinated nouns that cannot > take either a singular determiner or a plural determiner (although they > happily take a determiner which is indeterminate as to plurality). Examples: > (5) ?*This/ *These/ The book and magazine are quite old. > *This/ *These/ The idea and presentation are both quite boring. > *This/ *These/ The apple and orange are good to eat. > ?*This/ *These/ The computer and screwdriver are for sale. > *This/ *These/ The story and movie are quite different. > (I would say something "semanticky" is going on here--e.g. "This apple and > orange are good to eat together" sounds much better to me.) You bet something semanticky is going on here. It's called real language. :-) Now, the main problem with what you're trying to do with the "structure" is that the traditional transformation representation of structure is totally inadequate to describe what's really going on inside someone's head. There is no way to differentiate between an item's form and its function, between what it looks like and how it is being used. Case in point. You claim [noun "and" noun] is plural by virtue of its agreement with the verb. I prefer to think that the *form* of it is plural, but that its *function* can be either singular or plural, and can in fact be both singular and plural in the same sentence. Whether a given sentence sounds okay to us depends on two things: have we heard something like it before, and can we make sense of it? (The two are not unrelated.) English is not too particular about whether the noun phrase functioning as the subject of the sentence is singular or plural: This kind of people are always complaining. A majority are in favor of it. I'm sure you can come up with scads of examples. (As an interesting sidelight, in Greek, neuter plural subjects almost always take a singular verb: "The trees is green". The more I think about it, the more I think 95% of all syntax is idiom. We speak by analogy with what we have heard, not by doing 50 layers of transformation on a linear string of symbols.) With regard to the "this" sentences in (5), they are wrong only in their semantic isolation. Whether or not they can be made to sound right depends entirely on whether you can give them a context in which "this" can perform its proper function, namely deixis, or "pointing out". (For those of you listening in, deixis is just the Greek work for 'pointing'.) In other words, the nouns A and B must be in a context where they can be used to indicate the basic unity of A and B. From (5) above: *This apple and orange are good to eat. but, with proper mental context: See these two piles of fruit? This apple and orange are good to eat, but that apple and orange are not. Note that there need be nothing inherent in the relationship between noun A and noun B--the sentence can always *create* the relationship of unity: This telephone and baboon go together. This A and B go together. So to answer your first question: there's no reason A and B can't be coordinated. In fact, the deictic nature of "this" requires you to consider them coordinated, at some level or other. > Second question: why the difference between the two sets--those that take > singular determiners when coordinated (3), and those that don't (5)? Is it > something to do with whether the thing represented by the coordinate nouns > somehow forms a unit (in some gestalt(?) sense)? (If so, we should expect > quite a bit of difference in acceptability judgements about the examples in > (3) and (5).) Further examples of NPs like those in (3) and (5) are welcome. That's right on the money, by my lights. > Third question: if we have indeed ruled out the structure in (4), then it > would seem the only alternative structure is the following: > (6) [ [ Det N] "and" [ N]] > NP N? N? > (I put the question marks after the N labels on the inner brackets so as to > not pin myself down on the question of the number of bar-levels in NP.) If > this is the correct structure (and I welcome alternatives), then how does > the determiner get associated with the second noun? (I take it as clear that > the second noun in the examples of (2-3) is within the scope of the determiner.)Some kind of scoping/ movement rule, like with quantifiers? Rules, shmules. We say it like that because we're used to hearing it like that. Neurons are too slow to waste on a rule-based von Neuman approach. What I'm interested in is the mixed NPs: a) This hammer and screwdriver are mine. b) ?These nails and glue are mine. c) *This nails and glue are mine. d) ?This hammer and nails are mine. e) *?These hammer and nails are mine. f) *These hammer and screwdriver are mine. The fact that (b) and (d) are at all acceptable, and (c) and (e) are not, I attribute to analogical influence with the uncoordinated phrases "These nails" and "This hammer". Let's see, what's the rule that says things which are closer tend to govern? (e) is ever so slightly acceptable because of analogy with "these nails". (f) is a double loser in terms of analogy. (a) is a double winner, and that may be why we use (a) instead of (f). Note that this is basically a syntactic argument, and as such more or less unrelated with the semantic argument I proffered earlier. The syntactic argument explains why (a) sounds like something we've heard before. The semantic argument explains how we can make sense of (a). Does this sound like something you've heard before? :-) Does this make sense? :-) Larry Wall {allegra,burdvax,cbosgd,hplabs,ihnp4,sdcsvax}!sdcrdcf!lwall
friesen@psivax.UUCP (Stanley Friesen) (07/25/85)
In article <2188@sdcrdcf.UUCP> lwall@sdcrdcf.UUCP (Larry Wall) writes: > >You bet something semanticky is going on here. It's called real language. :-) >Now, the main problem with what you're trying to do with the "structure" is >that the traditional transformation representation of structure is totally >inadequate to describe what's really going on inside someone's head. There >is no way to differentiate between an item's form and its function, between >what it looks like and how it is being used. > This is why I prefer non-transformational approaches to grammer, such a Simon K. Dik's Relational Grammer. Unfortunately, even recent textbooks are still enamored with Chomsky and continue to present Transformational Grammer as the only approach, completely ignoring the other, more natural forms. -- Sarima (Stanley Friesen) {trwrb|allegra|cbosgd|hplabs|ihnp4|aero!uscvax!akgua}!sdcrdcf!psivax!friesen or {ttdica|quad1|bellcore|scgvaxd}!psivax!friesen
mmar@sphinx.UChicago.UUCP (Mitchell Marks) (07/29/85)
Those who have been following this discussion might enjoy a look at Jerry
Sadock's article "The Necessary Overlapping of Grammatical Components" in
the CLS 19 Parassession (Interplay of Phonology, Morphology, and Syntax).
One of his main examples is number agreement. He argues throughout the article
that the traditional hard walls between components of a grammar need more
'give', and cites examples of phenomena that are sometimes controlled from
one domain and sometimes another. Number agreement is offered as an overlap
of syntax and semantics.
This is not actually too far from Larry Wall's position as recently
posted, but without Wall's fierce anti-formalist dismissal.
--
-- Mitch Marks @ UChicago
...ihnp4!gargoyle!sphinx!mmarmichaelm@bcsaic.UUCP (michael b maxwell) (07/31/85)
Let me try to reply to/ comment on some of the replies/ comments so far about the question of why "this man and woman" behaves like a plural NP wrt number agreement on the verb, but takes a singular determiner (this). First, <498@mmintl.UUCP> franka@mmintl.UUCP> (Frank Adams) comments that some of the NPs that I marked as ungrammatical, e.g.: >(5) ?*This/ *These/ The book and magazine are quite old. > *This/ *These/ The idea and presentation are both quite boring. > *This/ *These/ The apple and orange are good to eat. > ?*This/ *These/ The computer and screwdriver are for sale. > *This/ *These/ The story and movie are quite different. --sound quite fine to him. I think he's right that they aren't really *ungrammatical* (no visual pun intended!), but it seems to me (and to some other people that I've checked with) that they aren't quite as *natural* as e.g. "this man and woman." I suggested further down in my original posting that this might have to do with whether the coordinated nouns form some sort of a "gestalt" unit. In another reply <2188@sdcrdcf.UUCP> from lwall@sdcrdcf.UUCP, Larry Wall offers several examples of CONTEXTS where some of the coordinate NPs in the list above sound quite fine. I agree, and I think something interesting is going on there. As he says, >the nouns A and B must be in a context where they can be used to >indicate the basic unity of A and B. However, I don't agree that this implies in *any way* that "the traditional transformation representation of structure is totally inadequate to describe what's really going on inside someone's head." First, let me say that I'm a generative grammarian, but not a transformationalist. By virtue of being a generativist, Larry's comments definitely apply to me (i.e. I think that there are definite syntactic facts about language, apart from any semantic facts). At the risk of getting into a long burning digression, let me offer a simple example where analogy fails completely to describe language, and a syntactic explanation is obviously :-) needed. In English, the word "that" when used as a sentence introducer (COMP) is generally optional, at least after a certain class of verbs. I think (that) John left. Furthermore, you can wh-extract virtually any argument of a verb--subject, object, locative, etc: Who do you think _ left? Who do you think John saw _? Where do you think John put the car _? Who do you think Mary gave the book to _? Finally, in most cases you can wh-extract any argument regardless of whether the word "that" appears: Who do you think that John saw _? Where do you think that John put the car _? Who do you think that Mary gave the book to _? By analogy with these sentences, you can *obviously* wh-extract the subject when the word "that" appears. WRONG: *Who do you think that _ left? We have heard something like it before, and we know exactly what it would mean if it were grammatical, but it isn't! This can only be syntax (well, at least it can't be semantics). By the way, this phenomenon isn't restricted to the complementizer "that," nor is it restricted to English... although it doesn't pertain to a *syntactically* characterizable set of languages (French is like English, Spanish is not, etc.) There are a host of similar examples in language, which a reading of recent (past 25 years, especially past 10 years) of generative syntax literature will reveal. As for Larry's comment-- >Rules, shmules. We say it like that because we're used to hearing it like >that. Neurons are too slow to waste on a rule-based von Neuman approach. --there's nothing inherantly von Neuman about phrase structure or rules. I think that what the mixing of syntax and semantics etc. shows (as proposed by Chomsky, incidentally) is that there is a large set of mental abilities we have, all of which interact to determine which sentences (phrases) we find acceptable; syntax is one, and autonomous in the sense that properties of syntax are not derivable from properties of other systems. Back to coordinate NPs. In <649@rlvd.UUCP> drg@rlvd.UUCP (Duncan R. Gibson) suggests that >This/These x and y are ... is an "abbreviation" for >This x is ... *and* this y is ... At first, I liked the idea; but consider the following: This man and woman met last week. *This man met last week and this woman met last week. This sort of sentence is an example of why linguists no longer try to derive (transformationally) coordinated NPs from coordinated Ss. So unless you can define "abbreviation for" in some other way than "derived from", this won't work. In <231@bbncc5.UUCP>, keesan@bbncc5.UUCP (Morris M. Keesan) suggests that there is an elided article in the relevant NPs, e.g.: > This book and [this] magazine are quite old. > The story and [the] movie are quite different. This sounds quite plausible; at the moment I can't think of any arguments which would distinguish between this analysis and one that says that "this" or "that" is a sort of quantifier which has scope over both nouns in logical form. Can anyone else? Later in his message, Larry Wall suggests that the explanation for the fact that the determiner must be singular in-- >> (2) This/ *These king and queen... --is that: >there may be some "pressure" exerted by the fact that the listener who has >heard "These king and queen..." has already analyzed "king" and "queen" as >nouns functioning as adjectives, as in "These king and queen mattresses...". >In other words, if "these A and B" were admitted to the language there >would be much more possibility of "garden pathalogical" noun phrases. I'm not sure I understand this; "these A and B" is quite fine if at least A (and preferably B, too) is plural: these men and women And I've seldom heard "man" or "woman" used as an adjective or as the first part of a compound noun, so it's not clear to me how an adjectival interpretation would arise in "these man and woman" that would lead me astray... But it may be that I just don't understand what he means. -- Mike Maxwell